Reviewers play a significant role in academic peer review activities, including conference paper assignment and funding selection, because their evaluation of proposals impacts the final decision. Several studies have proposed reviewer selection strategies or reviewer evaluation methods for solving the problem of selecting appropriate reviewers. Identifying reviewers who are familiar with the proposals to be reviewed is the objective of the reviewer assignment problem. However, the majority of the existing studies ignore quantitative constraints with respect to the articles assigned to the reviewers during the review process. In this study, we propose a novel optimization model with several review condition constraints to address the reviewer assignment problem. In the proposed model, the expertise and research areas of the candidate reviewers and proposals are identified using semantic topic models, which are demonstrated to be effective when measuring the relevance of the reviewers with respect to the proposals to be reviewed; further, the computational efficiency is improved owing to the reduced representation dimensionality. Herein, an improved heuristic algorithm is proposed to match reviewers and papers based on specific topic areas, and candidate reviewers are assigned to each proposal under the global optimum condition based on their overall performance values. Subsequently, an empirical test is conducted using a conference reviewer dataset. The obtained results show that the proposed model can help the managers to efficiently and effectively select reviewers in terms of the convergence rate and convergence level when compared with several classic benchmarks.

评审员在学术同行评审活动中起着重要作用，包括会议论文分配和资金选择，因为他们对提案的评估会影响最终决策。一些研究提出了审稿人选择策略或审稿人评价方法，以解决选择合适的审稿人的问题。确定熟悉要审核提案的审核员是审核员分配问题的目标。但是，大多数现有研究忽略了在审阅过程中分配给审阅者的文章的数量限制。在这项研究中，我们提出了一种具有多个审核条件约束的新颖的优化模型，以解决审核者分配问题。在建议的模型中，使用语义主题模型确定候选审稿人和提议的专业知识和研究领域，这在衡量审稿人与待审提议的相关性时被证明是有效的；此外，由于减小了表示维数，因此提高了计算效率。在本文中，提出了一种改进的启发式算法，以基于特定主题领域来匹配审稿人和论文，并在全局最佳条件下根据其总体绩效值将候选审稿人分配给每个提案。随后，使用会议审阅者数据集进行了实证检验。